Versions Compared

Old Version 12

changes.mady.by.user Christine Connolly

Saved on

compared with

New Version Current

changes.mady.by.user Christine Connolly

Saved on

Key

  • This line was added.
  • This line was removed.
  • Formatting was changed.

The Findings About Events or Interventions structure, referred to as the FA Structure in this section, represents collected data about an event or intervention that cannot be represented within an event or intervention record or as a supplemental qualifier to such a record. For example, this may be the case when:

  • There are several items which may be grouped together. If so, the FA structure allows the use of FAGRPID, FACAT, or FASCAT to group the items.
  • The observation is best represented in a Findings general observation class structure. If so, the FA structure allows the use of FAORRES, FAORRESU, and FAMETHOD for results, units, and methods respectively.
  • There are multiple evaluators. If so, the FA structure allows the use of FAEVALID.  

The variable --OBJ is unique to the FA Structure and is used with FATESTCD to represent what the topic of the observation is. FATESTCD describes the measurement/evaluation and FAOBJ describes the event or intervention that the measurement/evaluation is about. When collected data will be represented in a qualifier variable and are represented in the FA domain, the name of the variable will be used as the value of FATESTCD (e.g., FATESTCD = "OCCUR" and FATEST = "Occurrence Indicator"). The use of the same names (e.g., OCCUR) for both qualifier variables in the observation classes and FATESTCD is deliberate, but should not lead implementers to conclude that the collection of such data (e.g., occurrence) must be stored in the FA domain. In fact, data should only be stored in the FA domain if they do not fit in the general observation-class domain. If the data describe the underlying event or intervention as a whole and share its timing, then the data should be stored as a qualifier of the general observation-class events or interventions record. A record in FA may or may not have a parent record in an events Events or interventions Interventions domain. If an FA record does have a parent record, the value in FAOBJ should match the value in --TERM or --TRT, unless the parent domain is dictionary coded or subject to controlled terminology. In such cases, the value represented in which case FAOBJ should will match the value represented in --DECOD.

When to Use the FA Structure

...

Metadataspec
NumCriteriaDescription
1Collected observations have a timing that The timing of collected observations is different from an the timing of the associated event or intervention as a whole.
  • If the data item represents some action during or after the event or intervention, it may be considered to have its own timing, and meet Criterion 1
  • This criterion is less likely to apply to interventions records than to events records. 
  • A finding that is about part of an event, rather than the event as a whole, meets this criterion for the use of FA. An assessment of an event that is not about the whole of an event may be a " snapshot, " an assessment at a point in time, or a " slice," an  an assessment over a period of time (i.e., evaluation interval) during the event.
  • Assessments of parts of events (snapshots or slices) are represented in FA and may or may not have parent records.
    • If the FA dataset is split by parent domain, the applicant will decide which Events domain would have held a parent record for a parent-less FA record.
2Collected observations require more than 1 variable for representation.This may be the case when:
  • There several items which may be grouped together. If so, the FA structure allows the use of FAGRPID, FACAT, or FASCAT to group the items.
  • The observation is best represented in a Findings general observation class structure. If so, the FA structure allows the use of FAORRES, FAORRESU, and FAMETHOD for results, units, and methods respectively.
    • There are multiple evaluators. If so, the FA structure allows the use of FAEVALID.  

    Representation of a complex quality of an observation requires 2 or more related variables not in the events or interventions classes.

    • The need to represent data which require more than 1 variable in a findings about structure, rather than by adding 2 or more supplemental qualifiers to an Events or Interventions domain, is driven by the fact that each supplemental qualifier is in a separate record that links only to the parent record.
    • For example, a finding about an event that has a result with an associated unit. 
    3Collected observations indicate represent the occurrence of pre-specified prespecified adverse events.
    • Every record in the AE domain must represent an event that actually occurred. Therefore, AE probing questions that are answered in the negative (e.g., did not occur, unknown, not done) cannot be stored in the AE domain.
    • All answers to probing questions about the occurrence of pre-specified prespecified adverse events (e.g., "Y", "N", or "NOT DONE") will be stored in the FA domain.
    • For each "Y" response to a probing question there will be a record in the AE domain.
    • The FA record and the AE record will be linked via RELREC. 

    Naming Findings About DomainsCreating Datasets

    Applicants may choose to represent data in a single FA dataset (potentially splitting the FA domain into physically separate datasets following the guidance described in Section 4.1.6, Additional Guidance on Dataset Naming), or separate datasets, assigning , split FA datasets, or in separate datasets with unique custom 2-character domain codes (see examples in Section 6.4.5, Skin Response).

    For example, if findings about clinical events and findings about medical history are collected in a study, these could be represented as:

    ...

    .

    Metadataspec
    NumRepresentationImplementation
    1Single FA domain
    • A single FA dataset
    • Observations will be grouped as applicable by values in FACAT and/or FASCAT

    ...

    • The DOMAIN value would be “FA”.
    • Variables that require a prefix would use “FA”.
    • The dataset names would be the domain name plus up to 2 additional characters indicating the parent domain (e.g., FACE for Findings About Clinical Events, FAMH for Findings About Medical History). This naming convention may be used for an FA domain that has a parent domain even when the study has only 1 FA dataset that is not being split.
    • FASEQ must be unique within USUBJID for all records across the split datasets.
    • Supplemental qualifier datasets would need to be managed at the split-file level (e.g., suppface.xpt, suppfamh.xpt). Within each supplemental qualifier dataset, RDOMAIN would be "FA".
    • If a dataset-level RELREC is defined (e.g., between the CE and FACE datasets), then RDOMAIN may contain up to 4 characters to effectively describe the relationship between the CE parent records and the FACE child records.
    • .
    2Split FA domain
    3Separate domains
    • Separate datasets

    ...

    • where:
      • The DOMAIN value is

    ...

      • applicant-defined and does not begin with FA

    ...

      • .
      • All

    ...

      • guidance for the FA Structure is adhered to.
        • The --OBJ variable cannot be added to a standard Findings domain. A domain is either a Findings domain or a Findings About domain, not one or the other depending on the situation.

    ...

        • When the --OBJ variable is included in a domain, this identifies it as an FA domain

    ...

        • .

    ...

    • Variables that require a prefix would use the 2-character domain code chosen.

    Pagenav
    For the naming of datasets with findings about events or interventions for associated persons, refer to the SDTMIG: Associated Persons (available at https://www.cdisc.org/standards/foundational/sdtm).